We run additional empirical exercises to validate propositions in the framework identifying channels through which PPPs benefit the overall economy. For this, we follow the specification adopted by Cerra et al. (2016) identifying the factors explaining differences in the levels and quality of infrastructure in Latin American and Caribbean countries. The specification is as follows:
| Infrait = α + Χ′itβ + εit'′ | (2) |
| εit = αt + μi + Vit′ |
where Infrait are measures of infrastructure access and quality including
(i) access to electricity (% of total population and % of rural population),
(ii) telephone subscriptions per 100 people, (iii) mobile subscriptions per 100 people, (iv) improved water source (% of total population and % of rural population), (v) improved water sanitation (% of total population and % of rural population), (vi) road quality score, and (vii) overall infrastructure quality score. The column vector of independent variables is given by Χit. We are interested in the variable PPP investment as percentage of GDP to determine how PPPs potentially affect both access and quality of infrastructure services, which are identified as channels through which PPPs can deliver macroeconomic benefits. The disturbance term, εit, has orthogonal components: the country and year fixed effects, μi and αt, and the idiosyncratic shocks, Vit.
We also analyze variations in employment that may be associated with fluctuations in PPP investments. Employment created in the delivery of PPP projects could directly affect poverty, and Gutierrez et al. (2007) note that sector productivity and employment patterns may be important for reducing poverty. For this, we estimate the following equation with employment-share growth in major economic sectors as the dependent variable:
| Employit = α + Χ′itβ + εit'′ | (3) |
| εit = αt + μi +Vit′ |
The column vector of independent variables, Χit, include PPP investment (% of GDP), real per capita GDP, education expenditure (% of GDP), primary completion rate, minimum wage, and share of the working-age population; εit denotes the disturbance term.